Clonotypic IgH Response against Systemic Viral infection in Pronephros and Spleen of a Teleost Fish

The immune system is highly compartmentalized (1). Lymphocytes differentiate in primary lymphoid organs, circulate via the blood stream, and accumulate in secondary lymphoid organs. Although the frequency of lymphocytes specific for a given epitope is very low, secondary tissues constitute favorable microenvironments where Ags and APCs concentrate, thus increasing the probability of Ag encounter by specific B and T cells and initiation of clonal responses. In mammals, these secondary organs encapsulate highly organized tissues such as lymph nodes, spleen, as well as mucosal-associated structures such as tonsils and Peyer’s patches. In fish, the compartmentalization may be less apparent, because these species lack lymph nodes and a distinct regionalization of the spleen or mucosal lymphoid tissues (2). Also, sites where Ag-specific immune responses are initiated are not fully understood.

In mice and humans, genomic rearrangements of Ig loci occur in developing B cells in the bone marrow, leading to the expression of the BCR at the cell membrane (3). Newly formed immature B cells migrate from the bone marrow to the spleen after a first tolerance checkpoint and enter follicles where they persist for months in a resting state and become mature B cells ready to respond to their cognate Ag. T cell–dependent responses lead to particular structures named germinal centers, in which somatic hypermutation of Ig genes occurs, allowing selection of B cells expressing higher affinity Abs and a maturation of the response. Terminally differentiated plasma cells return to the bone marrow where they can be sustained for long periods of time and produce a large fraction of serum IgG (3, 4). Extrafollicular B cell subsets comprising mouse B1 cells, which are primarily located in peritoneum and other serosa, and marginal zone B cells express a particular Ig repertoire and follow distinct developmental pathways.

In fish, B lymphopoiesis occurs in the pronephros (or “head kidney”) (5–8). No surrogate L chain has been found in fish, and the mechanisms for allelic exclusion and for tolerance checkpoints are not known. However, a maturation gradient from the anterior (“head”) to posterior kidney has been shown in rainbow trout, leading to the idea that the head kidney is a primary lymphoid organ (similar to the bone marrow in mice and humans) whereas the posterior kidney would rather be a secondary lymphoid organ (similar to the spleen) (9). “B cell signatures” of key markers have been defined, combining Ig and stage-specific transcription factors, which distinguish early developing B cells, late developing B cells, and Ig-secreting cells (10). Importantly, the head kidney is also considered a niche for plasma cells, similar to the bone marrow in humans and mice (9, 10). Distinct repertoires of Ig-secreting cells responding to extract of the pathogen Vibrio anguillarum have been reported in rainbow trout spleen and head kidney (6).

In this work, we compared the modifications of head kidney and spleen IgHμ repertoires induced by a systemic viral infection with a rhabdovirus, the viral hemorrhagic septicemia virus (VHSV). We used a barcoded IgH 5′ RACE cDNA sequencing approach to characterize the whole IgHμ repertoires, that is, the VH gene usage, clonotypic composition (with a clonotype being defined as a given combination of a IGHV gene, a IGHJ gene, and a CDR3 amino acid sequence), expression, and clonotype sharing between individuals and immune status. This approach allowed to quantify the frequency of the antiviral public response in pronephros and spleen, across the whole IgHμ repertoire (i.e., taking into account all VH subgroups). Based on the comprehensive annotation of the rainbow trout genomic repertoire of Ig genes (11), we aimed at a gene-level description of expressed rearrangements. The public, neutralizing IgM response against VHSV (12–14) was present in both head kidney and spleen, and the frequency of anti-VHSV clonotypes evolved similarly in head kidney and spleen during the prime/boost protocol. However, amplified clonotypes were largely distinct between these organs, confirming that they express different B cell repertoires with limited overlap.

Rainbow trout isogenic line “B57” (gynogenetic, all females) (12, 13, 15) were raised in the fish facilities of the Institut National de la Recherche Agronomique (Jouy-en-Josas, France). All fish experiments were carried out in accordance with the recommendations of the European Union guidelines for the handling of laboratory animals (http://ec.europa.eu/environment/chemicals/lab_animals/index_en.htm). The experimental protocols were approved by the Institut National de la Recherche Agronomique Institutional Ethics Committee “Comethea” (permit license no. 15-60).

Two-year-old adult fish were immunized using the attenuated 25-111 variant of strain 07-71 of VHSV through i.m. injection at a dose of 10⁴ PFU/fish and were kept in 100 l tanks at 16°C. Primed fish were challenged 5 mo using the parental (nonattenuated) strain 07-71 of VHSV through i.m. injection at a dose of 10⁶ PFU/fish. This virulent strain allowed to check that fish were fully protected by the primary immunization. Infection by injection was selected to standardize the immunization and ensure that all fish received the same number of VHSV PFU. Sampling points of challenged and control groups are detailed in (Fig. 1. For sampling, fish were sacrificed by overexposure to 2-phenoxyethanol diluted at 1:1000. Blood was extracted and let to clot at 4°C overnight for serum extraction. Spleen and head kidney were removed and placed in RNAlater, then stored at −80°C until RNA preparation. Serum was extracted by centrifugation at 200 × g for 10 min, supernatants were collected and centrifuged at 1000 × g for 15 min, and serum was frozen at −80°C.

Sandwich ELISA was used for detection of total IgM levels in trout serum. ELISA plates were coated with 50 μl of 2 μg/ml mouse anti-trout IgM mAb 1.14 and incubated overnight at 4°C. Wells were then washed five times in 0.05% Tween 20 in PBS (T20-PBS) and the immunoassay was performed at room temperature as follows. Wells were blocked with 200 μl of 2% BSA in T20-PBS for 2 h. Serum samples were diluted 1:10,000 in T20-PBS with 2% BSA and 50 μl was added to the wells in duplicate. Plates were incubated for 1 h and washed five times before addition of 50 μl of biotinylated 1.14 mAb at 1 μg/ml. After a 30-min incubation, wells were washed five times and incubated with 50 μl of streptavidin-HRP (Thermo Scientific) at 0.2 μg/ml in T20-PBS with 2% BSA for another 30 min, washed five times, and the reaction was revealed using 100 μl of tetramethylbenzidine substrate (Sigma-Aldrich) in 0.05 M phosphate-citrate buffer (pH 5.0) with H₂O₂ for 30 min, protected from light. The reaction was stopped with 25 μl of 2 M H₂SO₄, and absorbance was measured in a microplate reader at 450 nm.

Indirect ELISA was used for detection of anti-VHSV specific IgM Abs in trout serum, based in published works for rabies virus detection (16, 17). The VHSV Ag used for the coating was a whole virus preparation of the strain 25-111, grown in epithelioma papulosum cyprini (EPCs) and purified by ultracentrifugation in a glycerol cushion. ELISA plates were coated with 50 μl of 2 μg/ml purified VHSV or purified inactivated VHSV in 0.05 M sodium carbonate-bicarbonate buffer (pH 9.6) overnight at 4°C. Optimal virus dilutions were determined in a previous titration assay using anti-VHSV Abs mouse monoclonal 192A17 and rabbit polyclonal 616. Washing and blocking steps were performed as described above. Trout serum was diluted at 1:200 in T20-PBS with 2% BSA. Biotinylated 1.14 mAb, streptavidin-HRP, and tetramethylbenzidine substrate were added as described above.

Virus neutralization assay with complement addition was performed in 24-well plates. EPCs were seeded into 24-well plates at 1 × 10⁶ cells per well in complete Glasgow’s minimum essential medium with 10% FBS and incubated overnight at 28°C. Trout serum was de-complemented and incubated at 10-fold serial dilution with VHSV (07.71 strain, 1 × 10² PFU/ml final) in complete Glasgow’s minimum essential medium with 0.5 mg/ml DEAE dextran and a 1:40 dilution of normal control trout serum (as a source of complement) overnight at 4°C. The media of the EPC plates were removed gently by aspiration, and 100 μl of serum-virus incubation solutions was added in duplicate. The plates were incubated for virus adsorption at 15°C for 1 h with rotatory shaking every 15 min. Monolayers were overlaid with 1 ml of 0.75% methylcellulose in medium 2% FBS and plates were incubated at 15°C for 4–5 d. Monolayers were fixed by the addition of 2 ml of 10% formol on top of the methylcellulose for 1 h, washed in tap water, and stained by addition of 1% crystal violet solution for 1 h. Plates were rinsed with tap water and PFU were counted manually. The neutralizing titer was calculated as the highest trout serum dilution causing a 50% reduction of the average number of plaques in control cultures inoculated with control trout serum, complement, and virus.

Libraries for Illumina deep sequencing were constructed as described in Magadan et al. (18), based on procedures reported in Vollmers et al. (19). One microgram of total kidney or spleen RNA was used for library preparation. Reverse transcription was performed according to the manufacturer’s instructions using a SMARTer RACE 5′ kit (Clontech). cDNA synthesis starts from the C region–specific primers for IgM (Cmu2, Supplemental Table I) and, with the template-switch effect, incorporates the universal adaptor (SMARTer II A oligonucleotide) at the 5′ end of the library. Second-strand synthesis was done using SeqAmp DNA polymerase (Clontech) (98°C for 3 min, 59°C for 4 min, 72°C for 10 min, 4°C) and primers containing Illumina adapter and a 15 random nucleotide (unique molecule identifier) for cDNA barcoding (1 μM Rd2p_UMI_5RACE oligonucleotide, Supplemental Table I). ds-cDNA was purified using Mag-Bind total pure next-generation sequencing beads (VWR) at a ratio of 1:1. The cDNA library was further amplified in a two-stage PCR. ds-cDNA was amplified with DreamTaq (Thermo Scientific) with 0.3 μM forward primers containing Illumina adapter sequences and six fish barcode (FBD) nucleotides (Rd2_FBD_Rd2p, Supplemental Table I) and reverse nested primer for IgM (Cmu1, see Supplemental Table I). Cycling was as follows: 95°C for 5 min, 30 times 95°C for 30 s, 59°C for 35 s, 72°C for 60 s, 72°C for 7 min, 4°C. The first PCR product was purified by gel purification (gel fragment between 500 and 800 bp) using NucleoSpin gel and a PCR clean-up kit (Clontech) and submitted to a second amplification with the same forward primer and a reverse primer with Illumina adapter and different numbers of random nucleotides at the 5′ end for different samples (Rd1_2N(4N)_Cmu1, see Supplemental Table I). This provides better diversity generation, which is critical for optimal accuracy of Illumina sequencing. Cycling was as follows: 95°C for 5 min, 15 times 95°C for 30 s, 62°C for 35 s, 72°C for 60 s, 72°C for 7 min, 4°C. The second PCR product was purified by gel purification using NucleoSpin gel and a PCR clean-up kit (Clontech).

RACE libraries were sequenced in paired-end 2 × 300-bp runs, using a MiSeq M01342 instrument (Illumina) and the MiSeq reagent kit v3 (600 cycles) (Illumina). Our consensus read sequencing approach based on the incorporation of a unique random barcode (unique identifier [UID]) in each cDNA molecule allows accurate quantification of clonotype frequencies and correction of PCR/sequencing errors. Sequences were filtered and merged using pRESTO, and consensuses were computed from read pairs and annotated using IMGT/HighV-QUEST, according to IMGT (http://www.imgt.org/) gene tables and the standardized IMGT nomenclature of IGH genes (11, 18). Each consensus was therefore produced based on the random barcode (UID), a IGHV gene, a C type (µ), an in-frame CDR3 sequence, and a J segment. Because we used an isogenic clone of rainbow trout in which the two haplotypes are quasi-identical, the annotation was likely easier and more reliable. UID and CDR3 sequences were combined, as previously described in Magadan et al. (13), to build a unique molecular identifier (MID) allowing a correction of PCR biases. The expression level of the clonotype, defined, for a given isotype, by a triplet (V gene, J gene, CDR3 sequence) was measured by counting the corresponding MID barcodes.

A DiversiTR Web-based interface (https://github.com/ph-pham/DiversiTR.) was used. Repertoire diversity was assessed by computing the Gini index, which measures the inequality of a clonotype distribution. It ranges between 0 and 1, with 0 corresponding to perfect equality (i.e., all clonotypes are represented at similar frequencies), and 1 representing perfect inequality (some clonotypes are more frequently observed than others). Differentially expressed VH genes between experimental groups were identified using the DESeq2 R package. Volcano plots were then plotted showing log₂ fold change (x-axis) and significance (−log₁₀ adjusted p value; y-axis) of differentially expressed genes. Significant differentially expressed clonotypes were determined when p < 0.05, and a log₂ fold change threshold ≥2. To select the top frequent 50 clonotypes shared across immunized fish, the expression of clonotypes across all immunized fish at all time points was summed from 10,000 MID subsamples and used to rank clonotypes. CDR3 length spectratypes were computed from sequences encoding productive rearrangements only, based on subsampled normalized data.

Sequence data used in this work are registered in the BioProject National Center for Biotechnology Information database with the SRA accession number PRJNA798218 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA798218/).

To investigate whether and how head kidney B cells are involved in responses against VSHV infection, modifications of the B cell repertoire were analyzed in this tissue after primary and secondary immunization. To this purpose, six experimental groups were established (Fig. 1). The primary B cell response and its persistence were investigated at 1 and 5 mo postimmunization with VHSV (groups Pr1m and Pr5m). The secondary response was analyzed 1 and 4 wk after a boost performed 5 mo after prime (groups Bst1w and Bst4w). Noninfected control groups were sampled 5 and 6 mo after the beginning of the experiment. Isogenic fish with the same genetic background as in our previous work (clone B57) were used, ensuring the best possibilities of repertoire comparison across individuals and studies (12, 13).

The Ab response induced by immunization, prime or boost, was first measured in the serum. All immunized groups showed increased total serum IgM titers compared with controls (Fig. 2A), as we previously observed (12), but no significant differences appeared between time points or conditions. In contrast, specific anti-VHSV Abs (Fig. 2B) were detected 1 mo after primary immunization (Pr1m), but their titers were significantly higher 5 mo later (Pr5m), also in line with previous reports (20). Surprisingly, serum anti-VHSV Abs titers decreased 1 wk after boost and even more 4 wk later. Neutralizing Ab titers (Fig. 2C) were significant in all immunized groups and did not increase significantly after boost. Hence, secondary immunization did not lead to stronger and faster production of Ag-specific/neutralizing Abs, as compared with that obtained after primary immunization.

To characterize modifications of the head kidney or spleen B cell repertoire induced by the response to VHSV, we generated isotype-specific Ig repertoire libraries using a cDNA 5′ RACE–based protocol as in Magadan et al. (18) and Egorov et al. (21). Defining clonotypes as triplets (IGHV gene, IGHJ gene, CDR3 [amino acid] sequence), we assessed their expression level from subsampled datasets based on cDNA unique MIDs as established in Magadan et al. (13, 18).

The overall IgM clonotypic diversity did not differ significantly from one condition to the other with the exception of some individuals in the group Pr1m, in the head kidney (Fig. 3A) as well as in the spleen (Fig. 3B). These individuals show an increased proportion of “larger” (i.e., more frequent) clonotypes early after primary immunization; this is also shown by the variations of the Gini index reflecting the inequality of clonotype expression (Supplemental Fig. 1).

Differential analysis of the expression of VH/JH distributions between groups did not identify highly responding combinations, as no significant increase of frequency was observed after immunization for any VH/JH combination. However, several combinations tended to be consistently more frequent in the head kidney of all individuals of the Pr1m group (fold change ≥ 1), involving IGHV genes 1-18, 1-21, 4D-23/24, 4D-43, 8-5, and 8-19 (Fig. 4A).

We then compared IGHμ CDR3 length distributions between controls and fish after primary immunization (Pr1m group) (Fig. 4B). Interestingly, for VH genes indicated by the differential analysis such as IGVH1-18 and IGHV4D-23, most spectratypes of individuals from the Pr1m group were clearly perturbated compared with controls, with multiple clonal expansions of convergent CDR3 length induced by immunization. Taken together, these observations point to an IgM response involving multiple VH genes 1 mo after priming, and decreasing 4 mo later.

The strength of the IgM response in the head kidney was also illustrated by the proportion of the most expressed clonotypes in each fish, classified by ranges of rank (Figs. 3C, 4C). Strongest responses were detected in the Pr1m group again, with the 100 most frequent clonotypes corresponding to 20% of the expressed repertoire or more in two individuals. The other immunized groups showed much lower contribution of the top frequent clonotypes. A similar pattern was observed in the spleen.

We previously identified a strong public response against VHSV, present in the spleen of all immunized B57 isogenic trout (12). This response involved IgHμ clonotypes expressing IGHV1-18 and IGHJ3 genes (previously known as VH5 and JH5; see Ref. 11 for nomenclature issues), rearranged into eight highly similar CDR3 sequences with 10 aa (13). In these previous studies, IgH clonotypes were amplified with VH subgroup-specific primers; hence, we were not able to calculate the frequency of this public response. In the present work, the approach based on 5′ RACE libraries allowed us to determine accurately the frequency of clonotypes in the whole repertoire, and also to look for additional public responses that might involve VH subgroups we had not sequenced previously. To get insights into the response against the virus, we selected from normalized data (subsamples) the top frequent 50 clonotypes shared across immunized groups (see Materials and Methods) and analyzed their expression across all individuals. These top clonotypes could be classified into four sets, as described below in (Fig. 5.

These clonotypes were >8-fold more frequent in immunized fish compared with controls. These eight clonotypes correspond precisely to the IGHV1-18/IGHJ3 public response we previously described, with a 10-aa CDR3 (12–14).

Seventeen of these 18 clonotypes were found in the Pr1m group, and in at least one of the other immunized groups. Three clonotypes expressed IGHV9-23, and one expressed IGHV9-15 in combination with different J genes. Two clonotypes expressed IGHV6-4 and IGHV6-31, respectively. The others expressed IGHV1 genes: three IGHV1-13; five IGHV1-18; and one IGHV1-39, IGHV-41, and IGHV-42. Finally, two clonotypes shared the same CDR3 sequence ARSNNGGAFDY, associated either to IGHV1-18 or IGHV1-42 and IGHJ3. CDR3 sequences were of variable length (9–13 aa). Most of these clonotypes were not seen in controls. Clonotypes in set 2 were >8-fold more frequent in immunized fish than in controls.

These sequences were not found in controls, and most of them (17 out of 20) were seen in the Pr1m group only. In absence of systematic correlation between clonal expansion and viral immunization, these clonotypes may correspond either to clones specifically induced by the VHSV infection, or to responses that occurred prior to immunization. However, their observation quasi-exclusive 1 mo after prime suggests that collectively they were rather induced by immunization, either specifically recognizing VHSV or due to bystander activation. This pattern is also coherent with a reduction of the diversity of the anti-VHSV response between Pr1m and Pr5m.

These four clonotypes likely correspond to rearrangements highly frequent both in controls and after immunization across all groups, possibly specific for prominent Ag present in the environment and not directed to VHSV epitopes. They express IGHV genes from subgroups 1 or 6.

These data confirm that the IGHV1-18/IGHJ3 public response against VHSV is very dominant. They also point toward a number of expanded clonotypes expressing VH from subgroups 1, 6, and 9, shared by several immunized fish. The expression pattern of these shared top clonotypes underlines a “convergent” response, that is, a reduction of diversity of the shared response with time and after boost. The higher diversity of the primary response is also illustrated by the expression of clonotypes by decreasing rank (Fig. 4C).

We then focused on these dominant public clonotypes to follow the evolution of the anti-VHSV IgM response across the prime/boost.

Overall, public clonotypes were consistently much more frequent in the head kidney of immunized fish than in controls (Table I). The response was high 1 mo after prime and was still detectable 4 mo later, as we previously observed in the spleen (13). Surprisingly, the frequency of anti-VHSV public clonotypes was clearly reduced after boost, although it remained much higher than in controls. This was observed as early as 1 wk after boost, a trend confirmed 3 wk later (Fig. 5, set 1, Table I). Importantly, Fig. 4C shows that these public responses were not replaced by large alternative private clonotypic expansions, suggesting an overall decrease of the VHSV response in head kidney after boost, compared with Pr1m or Pr5m fish.

We also looked for alternative clonotypic expansions expressing IGHV1-18 in immunized fish. We first analyzed the CDR3 length spectratypes reflecting all IGHV1-18 IgHμ transcripts (not only those expressing IGHJ3). As shown in (Fig. 6A, mainly the 10-aa peak corresponding to public clonotypes was enhanced after immunization. For example, perturbated profiles appeared in three fish out of four in the Pr1m group. Interestingly, expanded peaks at 10 aa were also observed in IGHV1-18 CDR3 length profiles of the group Bs1w (fish nos. 17, 18, and especially 19). However, these expansions did not correspond to B cells expressing public clonotypes but, in this case, by the increased frequency of a clonotype annotated IGHV1-18/IGHJ2D (Fig. 6B). This clonotype was amplified in three fish out of four in the Bs1w group, but it was not detected in any other immunized individuals. Thus, this “transitory” public expansion had disappeared at 1 mo postboost in the Bst4w group.

To compare VHSV-induced modifications of the IgM repertoire between head kidney and spleen, we then analyzed the spleen IgM repertoire in the control group Ctrl5 and in all immunized groups, using the same approach of 5′ RACE RepSeq. The IgHμ repertoire was sequenced in three fish per group, except in Pr1m, in which all four individuals were analyzed.

To compare the intensity of the IgHμ repertoire modifications in these tissues, we first focused on the IGHV1-18/JH3 public response because it was clearly specific for VHSV. We found no clear difference (Table II): the public response was present in both spleen and head kidney, at variable frequencies within immunized groups. There was no significant difference between the two organs and no obvious trend across groups.

We then extended our analysis to all IGHV1-18/Cμ rearrangements, comparing CDR3 length spectratypes from head kidney (Fig. 6A) and spleen (Fig. 6C). We observed, for most individual fish rather similar spectratypes, suggesting that the intensity of the response evolved in a parallel manner in these two tissues.

We also represented in parallel the frequencies of the IGHV1-IGJH3 public clonotypes and of the most frequent IgHμ clonotypes (Fig. 7). The intensity of the public response was variable from fish to fish, but not obviously different between head kidney and spleen.

We next asked whether head kidney and spleen represented distinct B cell compartments or, whether alternatively, responding clones freely circulated between these organs, hence showing correlated frequencies. The most frequent clonotypes (>5.10⁻⁵ or >15.10⁻⁴) were selected for each fish from head kidney and spleen, and the level of sharing between these tissues was assessed in each individual (Fig. 8A, 8B). To test whether cells expressing the same IgH VDJ rearrangement were present in head kidney and/or spleen, we considered clonotypes defined at the nucleotide level. Strikingly, there was a very limited overlap between the two organs, not only in controls but also after immunization. Hence, B cells present in these tissues constituted distinct compartments because the most frequent, that is, potentially responding, B cell clones mostly expressed different rearrangements.

We also checked whether top clonotypes shared between head kidney of different individuals were also shared between head kidney and spleen. We therefore focused on sets 1, 2, and 3 defined in (Fig. 5. Public clonotypes (from set 1) were generally detected in both tissues in each fish. In contrast, top clonotypes from the other sets were mostly found either in head kidney or spleen, even the most frequent ones, supporting the hypothesis of two distinct compartments (Fig. 8C)

In teleost fish, the pronephros (or head kidney) encompasses a considerable amount of lymphoid tissue that plays roles similar to human and mouse bone marrow in hematopoiesis and lymphopoiesis (22). Although this tissue is likely important for production of serum Abs and likely also for immune memory, its contribution to B cell responses during infections or after vaccination remains largely unknown. In the present study, we investigated the involvement of the head kidney in the B cell response against a systemic viral infection, the VSHV virus, comparing the modifications of its IgHμ repertoire with that of a typical secondary lymphoid organ, the spleen, throughout a prime/boost protocol.

Although the VH usage in head kidney was not overall strongly affected by VHSV immunization, IgHμ repertoire analysis identified significant changes involving multiple clonotypes. Several VH subgroups were involved across time points, mainly subgroups 1 (genes IGHV1-13, IGHV1-18, IGHV1-39, IGHV1-31, and IGHV1-42), 6 (genes IGHV6-4 and IGHV6-31), and 9 (genes IGHV9-15 and IGHV9-23). Our 5′ RACE–based libraries were representative of the relative clonotype frequencies within the whole IgHμ repertoire. Indeed, we could therefore quantify the relative importance of different expanded rearrangements in the total response. Among the most amplified sequences, over all immunized groups in average, we selected the clonotypes that were detected in, and shared by, most immunized fish. Strikingly, all of these clonotypes expressed IGHV1-18 and IGHJH3 genes, and they corresponded to the public response to VHSV found in the spleen in our previous studies (12, 13). These observations reveal the presence of B cells responding to the virus in the head kidney at different time points after immunization, in line with the findings of Zwollo et al. (9, 10). The public response against the virus was based, at least partly, on the same small set of clonotypes in head kidney and spleen. Importantly, the frequency of this public response could be accurately quantified by our 5′ RACE–based approach, showing that the public IGHV1-18/IGHJ3 response was very dominant.

We also analyzed the modifications of the IgHμ repertoire in the spleen of the same individuals to investigate whether the response induced by the virus followed similar trends in head kidney and spleen. The amplitude of changes induced by prime and boost with VHSV was globally similar at all time points analyzed: a stronger and more diverse primary response was seen both in spleen as in the head kidney, and the IGHV1-18/IGHJ3 public response was dominant. These data confirmed our previous observations, in particular the persistence of expanded VHSV-specific clonotypes in the spleen 5 mo after primary immunization (13). However, comparing the head kidney and spleen repertoires of individual fish, we could not find any clear (positive or negative) correlation between individual expanded clonotypes in these tissues, neither for frequency of the public response nor for distribution of top frequent clonotypes in spleen versus head kidney (Figs. 7, 8). Previous reports have described distinct B cell repertoires in spleen and head kidney in naive fish or after hapten-carrier immunization (5, 6, 20). Furthermore, our data uncover that, during infection or after vaccination, these two compartments are both subjected to the effects of systemic response to immunization/infections, but they are not homogenized by fast cell recirculation.

Another new finding of our analysis is the reduced expression and reduced diversity of the anti-VHSV clonotypes after boost (see (Fig. 5, especially set 3). Focusing of the top 50 clonotypes detected in most immunized fish (i.e., sets 1, 2, and 3 in (Fig. 4), we observed a strong reduction of the number and diversity between the primary response and all the other groups, that is, 5 mo after priming and even more after secondary immunization. If particular clonotypes would be specifically expanded after secondary immunizations (but not in other time points), they might not be part of clonotype sets 1–3; for example, the IGHV1-18 clonotype ARYTGNAFDY was expanded in three fish 1 wk postboost but not in the other groups (Fig. 6B), and it was therefore not kept in the top list selected in sets 1–3. Importantly, the ranked distribution of clonotype frequencies (Fig. 4C) shows that such situations are rather exceptional. Hence, the primary response in the head kidney indeed recruits many more clonotypes and leads to stronger Ab repertoire modifications than observed later in our protocol.

It is then interesting to draw a parallel between these observations and the titer of anti-VHSV Abs measured in the serum. Titers more than doubled between 1 and 5 mo after primary immunization, as previously reported for other Ags (23). This was likely explained by the secretion of large amounts of Abs by VHSV-specific plasmablasts and plasma cells during the first months after prime. More intriguing is the decrease of serum Ab titers after boost, in the context of a lack of a clear secondary response. Secondary responses in fish appear to be strongly affected by the dose of the Ag during priming: fish immunized with a low Ag dose show typical secondary responses with enhanced titers after secondary immunization, whereas fish first injected with a high dose of Ag show a high primary response but no secondary response (24, 25). Immunization by VHSV i.m. injection leads to an acute systemic infection and should certainly be considered as an “Ag high dose” prime. However, the decrease of serum Ab titers and the reduction of diversity and frequency of anti-VHSV clonotypes, both in head kidney and spleen, raise several questions. Are B cells producing many top clonotypes observed in immunized fish after prime, besides the public response, mainly due to bystander activation, or are they specific for VHSV but not sustained during the months following primary immunization, perhaps because the remaining amount of Ag is not sufficient? Why then can B cells (plasmablasts/plasma cells?) producing public clonotypes persist for months after prime, and after secondary immunization, both in spleen and head kidney? As fish express neutralizing Abs at the time of the challenge, it is possible that the effective dose of virus available to stimulate B cells after neutralization could be rather low, hence eliciting a narrow response of B cells with lower stimulation threshold. This might explain the contraction of the anti-VHSV B cell repertoire suggested by our data after boost. Another factor determining the composition of head kidney and spleen IgH repertoire is the location of B cells. Might Ag-specific B cells (and/or plasmablasts, plasma cells) egress lymphoid tissues and relocate close to lesions, producing Abs that may not primarily accumulate in the serum? This hypothesis could be supported by observations of circulating B cells induced by salmonid alphavirus in Atlantic salmon, which leave the spleen and are recruited in the heart after challenge (26). Taken together, these observations challenge a model in which, as in humans and mice, fish serum Abs would be mainly produced by resident plasma cells located in the head kidney, the equivalent of the mouse bone marrow. If Ig-producing B cells are more prone to relocation close to lesions/infection sites during secondary responses (vascular endothelia in the case of VHSV), their protective action should be understood within a fully different model.

Our model of i.m. injection of the virus does not mimic the natural infection by VHSV, for which the fin bases are the major portal of entry (27). In our model, a strong systemic infection rapidly develops without stimulation of a mucosal/skin reaction. We are aware that Ag dose and route of administration might have a significant impact on head kidney and spleen B cell responses. Further experiments are required to understand how the clonal dynamics in spleen and head kidney can match the evolution of serum anti-VHSV titers after different types of immunizations.

We thank Louis du Pasquier and Oystein Evensen for insightful discussions about the data and Catherine Collins for comments on the manuscript.

The authors have no financial conflicts of interest.